After the refugee crisis: public discourse and policy change in Denmark, Norway and Sweden

The refugee crisis of 2015 can be characterised as an exogenous shock, an immigration shock, which according to institutional theory may create a window of opportunity for path-breaking policy change. In all three Scandinavian countries a range of new policy proposals were made in order to stem the incoming migration. Although the direction of policy change pointed in a similar restrictive direction, the choice of policy instruments differed, as did the style and content of policy actors’ legitimizing and coordinating discourses. Drawing on the analytical framework of discursive institutionalism, the three cases illustrate how a similar kind of external shock (the refugee crisis) was construed differently in different national public spheres, and generated different kinds of policy responses – in particular with respect to the social rights of refugees. The emphasis on legitimizing communicative discourse and coordinating discourse differed between the countries. In Sweden, the need to legitimize policy change normatively was acute. This mattered less in Denmark, where emphasis was on creating stable political support for a range of specific measures. Norway takes a mixed position between the two. Data includes a selection of media texts from six newspapers in 2015 as well as policy texts from 2015 and 2016.

Introduction 2015 was a dramatic year in the history of Europe. War and poverty led thousands of humans to cross the Mediterranean in ramshackle vessels. Many drowned. Even more ended up in Greek or Italian detention centres or moved on to countries further north. The events were highly mediatized. The visual images were strikingsome heart breaking, others frightening. Also in Scandinavia, the plight of the refugees, as well as the challenges involved in accommodating growing numbers of arrivals were hotly debated, and the dilemmas between handling international humanitarian crisis and protecting national welfare systems were brought to the fore.
It seems apt to characterize the situation in the fall of 2015 as one where national models of immigration control and refugee integration encountered an "exogenous shock". In literatures on policy change and resilience, such shocks and external pressures are associated with change. Windows of opportunity for change open up in otherwise stable and path-dependent policy models (Pierson, 1996), and external pressures generate change in similar directions across countries often taking the shape of a "race to the bottom". With thousands of refugees on the move, states strived to avoid becoming their preferred destination and consequently sought to offer less favourable conditions for entry and welfare. Hernes (2018) has shown that all the three Scandinavian countries changed policies in a restrictive direction following the so-called refugee crisis. This included stricter regulatory policies, but also restrictions in refugees' access to social rights. However, while the direction of change is similar, the countries chose different policy instruments to pursue their goals, thus showing traits of path-dependency. This observation of persistent national characteristics, also in times of exogenous pressure and a general move towards restrictive policies, is interesting in light of concerns that there is a race to the bottom taking place in immigration politics. But it also invites further reflections on what it is in the respective national contexts, and beyond, that works to produce nationally specific responses at a moment when so many call for international coordination, and where policy challenges are seemingly of a transnational nature.
Policymaking are sense-making activities (Campbell & Pedersen, 2014). When existing policies appear unable to deal with new situations and challenges, policy actors will strive to come up with new ideas, new policies that appear better suited to interpret and cope with the current developments. But before new policies are hammered out, policy actors as well as national publics need to make sense of the new situation itself. In this process, the media plays an important part through how the challenge is framed and presented in the public debate. While the transnational movements of people, the deaths in the Mediterranean and the devastating images of Alan Kurdi were the same, this study demonstrates that the crisis was construed in different ways in different placesboth in the media coverage and in policy discourse. In the words of Widmaier, Blyth, & Seabrooke: "even exogenous shocks must be interpreted" (2007, p. 748).
In this article I argue that migration studies need to pay attention to discursive processes of sense-making and policy construction in order to understand variation in immigration and integration policy. The cases of Sweden, Denmark and Norway demonstrate that the refugee crisis was construed differently in the three countries and that this had implications for policy choices. While all the countries raised the bar for asylum-seekers, they did so using different policy instruments. In particular, they varied with respect to how changes in refugees' social rights were applied as a tool in immigration policy. Drawing on theoretical insights from discursive institutionalism I demonstrate that this is not only a matter of variations in the content of discourse but also in the discursive work that goes into creating popular legitimacy (communicative discourse) and cross-political compromise over new policies (coordinative discourse). The study also contributes to this literature theoretically by showing how the emphasis on communicative and coordinative discursive work vary not only by political system (Schmidt, 2002) or stages in the policy process (Ervik & Lindén, 2015) but by the type of policy change that are targeted. Communicative discourse was most important in establishing new overarching goals for policymaking. When it came to transforming social policies to provide new policy instruments for migration control, coordinative discourse to establish broad parliamentary support was vital. For comparative migration studies, one implication is that different types of discursive processes are operative depending on whether it is strictly immigrant specific policies that are at stake, or if general welfare state institutions are reformed with the purpose of adapting to migration.
The research questions guiding my analysis have been the following: How was a similar set of eventsthe "refugee crisis"construed in the three Scandinavian countries? Furthermore, how did policy actors operate to change (and/or protect) existing policies with respect to refugees' social rights in light of the new situation, as it was construed in each national setting? In sum, the study contributes to a theoretical discussion about the role of problem construction and discursive processes within national public and political spheres in processes of policy change and resilience.
The article proceeds as follows: In the first part I outline an analytical approach to study the role of discourse in processes of policy change. This section also briefly reviews some existing studies of Scandinavian responses to the refugee crisis. I then outline the data used and analytical approach taken. In the main part of the article, I analyse the events of each country separately. Finally, I move on to discuss the cases in relation to each other and conclude.
Converging policies or resilient national models? A discursive approach Both in the migration and welfare state literature, there is an ongoing debate on the status of national models. Are similar exogenous pressures and transnational competition pushing once distinct models towards more uniform future policy regimes? Welfare scholars discuss the future of national welfare state regimes in times of globalization and permanent austerity (Pierson, 1996;Starke, 2006). Throughout Europe, concerns about the failure to integrate migrants into labour market and civic life have generated policy responses in the shape of integration courses and language tests, which resemble each other across seemingly distinct national models (Joppke, 2007). International human rights conventions and EU-regulations also push towards harmonization. Yet, some authors maintain that distinct national models can still be observed. Change happens, but it is path-dependent, thus maintaining national variation (Borevi, 2014;Goodman, 2010Goodman, , 2011Hernes, 2018). The institutionalist argument (Pierson, 1996) would be that sudden crisis or particularly strong external pressures can open up windows of opportunity for change even in stable institutions. This makes the 2015 refugee crisis and its aftermath a particularly intriguing case.
None of these contributions pay detailed attention to the smaller details of the politics of reform (see Boswell & Hampshire, 2017 for an exception). It is the force of existing institutions and national ideologies, or of external pressures, which explain resultant patterns. The focus tends to be on the resultant policies, more than on how they were reached at. The migration studies literature accordingly provide limited insight into how similar pressures and crises are interpreted and construed into distinct problems that require national policy responses.
The so-called discursive turn in policy studies emphasizes that in order to understand the dynamics of policy processes it is not enough to understand power constellations or historically evolved institutions. We must analyze how particular policy problems are construed and presented in ways that enable them to sway public opinion or forge political alliances (Bacchi, 2009;Béland & Cox, 2011;Schmidt, 2002;Stone, 2012). Vivien Schmidt's discursive institutionalism is useful in that she distinguishes between the ideational content of discourse and discourse as an interactive process where policy actors coordinate the construction of their policy programme and communicate it to the general public (Schmidt, 2002, p. 6). In other words, a study of discourse and policymaking must address both the content of what policy actors say and how they relate to each other, as well as to the general public through the mass media. Within the ideational dimension, Schmidt distinguishes between the cognitive and the normative dimension of discourse. The cognitive function elaborates the logic and necessity of a policy, while the normative function demonstrates its appropriateness by appealing to shared values. A successful discourse, in this perspective, tends to be one that succeeds in demonstrating both the functional necessity and normative desirability of change. The interactive dimension of discourse also has two functions, a coordinative and a communicative. Coordinative discourse takes place between policy actors and stakeholders. It is a type of discourse aimed at providing a common language and a shared framework for the construction of a policy programme. This is necessary to establish political coalitions that can carry through a reform. Communicative discourse is directed at the public. It relates to the public presentation of and deliberation over policy programmes, and seeks to establish popular legitimacy for a programme.
While I find the analytical framework developed by Schmidt useful, it seems somewhat underdeveloped with respect to the interlinkages between the mass media, public debate and policymaking. Schmidt focuses her attention at the discourse of policy actors, how they speak to each other and to the public. But their ideas do not appear from empty air; we also need to understand the political contexts and public spheres within which they construe their discourses. What works well to legitimize a policy in Denmark, may have opposite effects in Sweden.
This points to the centrality of problem definitions, −constructions, or frames. Complex situations, such as mass migration to Europe in 2015, can be defined differently depending both on which aspects are emphasized and on how they are presented. The efficiency of a particular problem definition will depend on a variety of factors, including the power and position of those who propagate it, as well as which metaphors and narratives that are employed, or which values that are made relevant (Bacchi, 2009;Mehta, 2011;Stone, 2012).
Policymaking in the wake of the refugee crisis in 2015 took place within wider, largely nationally confined, public discourses, where the media played an important role in framing the issue. The exact causal relationship between the problem definition processes that take place in the media and between policy actors is undoubtedly complex and multi-directionaland insufficiently theorized (Mehta, 2011). Previous research indicate that policy actors in the immigration field are highly conscious of media coverage and adapt their practices to this (Thorbjørnsrud, 2015), but also that journalistic framing is shaped by national cultural repertoires and institutions (Benson & Saguy, 2005). The point in this study is to focus on the process of discursive work that leads up to new policy: Policy does not simply reflect public discourse, but the public discourse constitutes a potentially important part of conditions policymakers operate within.
Different types of discursive work matter at different timesand policy actors prioritize their discursive efforts differently. In early 2015 the refugee crisis was predominantly taking place somewhere else. It was not an acute problem for national policy actors in Scandinavia. The media were more active than politicians in bringing the issues to the public agenda. But the national media framed the events in the Mediterranean in very different ways in the three countries. In the analysis below, I seek to link these early framings of the crisis in the media to an evolving policymaking process. As the refugees increasingly appeared on the borders policy actors had to produce solutions, but also their own problem definitions. Swedish, Danish and Norwegian policy actors (nearly) all aimed to reduce the arrival figures. They shared goals, but they legitimized and operationalized them differently. Furthermore, Schmidt's analytical distinction between the ideational aspects of discourse and its interactive functions is crucial. I identify not only variability in the content of discourse, but in the efforts governments put into establishing support for policies through discursive interaction with the public and other policy actors.

Immigration politics and refugees' social rights in Scandinavia
Scandinavia has traits that makes it a perfect site for a most-similar comparative case study. Despite the many similarities in size, culture, language, politics, labour markets and welfare systems, there exists a persistent pattern of intra-Scandinavian differences within immigration policy (Brochmann & Hagelund, 2012;Huddleston, Bilgili, Joki, & Vankova, 2015;Hernes, 2018). In this picture, Sweden has taken a comparatively liberal approach to immigration and cultural diversity. The country has received many refugees, newcomers' have equal social rights as citizens, and residence and citizenship policies have been very inclusive. Until very recently (also in this sense the refugee crisis represented a break) immigration policy has been relatively little contested in the political debate, and the country has taken pride in being a country with an open attitude to cultural diversity and strong stance against racism. Quite contrary to this, Denmark has adopted policies that are more restrictive and expresses a more pronounced ethnonational approach in dealing with immigration-induced diversity. There have been signs of welfare dualism, with reduced social benefits for new residents. The antiimmigration right-wing Danish People's Party (DPP) has been a support party for several conservative governments, but restrictive policies are by no means restricted to the right. Norway tends to be positioned somewhere between its two neighbours. It has never been as liberal as Sweden, but never ventured as far as Denmark in a restrictive direction. The anti-immigration Progress Party has been part of a coalition government since 2013, but appears as more moderate than the DPP.
More than 1,2 million first time asylum-seekers were registered in Europe in 2015. Also in the Scandinavian countries, the number of asylum seekers reached unprecedented heights. 156,110 (16016) first time asylum-seekers were registered in Sweden,30,470 (5898) in Norway, and 20,825 (3679) in Denmark (Eurostat, 2016, figures in parenthesis is number of applicants per million inhabitants). For Sweden and Norway, this constituted more than a doubling of 2014-figures. For Demark the growth was less dramatic, but still significant.
It is not surprising that the arrival of many refugees in a short period of time creates strains and tensions. Housing facilities and social services will be stretched, as will case handling capacities. Added to the logistical challenges is the politically contentious nature of contemporary refugee and immigration policy. Waves of sympathy and refugeefriendly grass roots initiatives were one aspect of this. But the growing influx also gave rise to anti-immigration sentiments and fears for the changes the sudden population movements would impose on receiving societies. Both in Norway, Sweden and Denmark, existing immigration control measures appeared as insufficient in the face of new and overwhelming migratory movements. Furthermore, questions were raised as to whether the existing welfare state apparatus would be able to handle the incoming migration, and thus if also existing social policies and the Nordic welfare model as such needed to reform.
The result (elaborated in more detail below) was one of both change and stability. Hernes ' (2018) analysis demonstrated that restrictive moves were made in all countries, but different policy instruments were chosen. She suggests that the mechanism of regulatory competition led to decline in goal variation: As all three countries sought to avoid appearing attractive to migrants, they entered the European "race to the bottom". But, she argues, the shape of existing policies affected also crisis responses generating a pattern of path dependency in the choice of policy instruments as these varied between the countries.
The present article connects not only to previous studies of policy development following the refugee crisis, but also to research on the public discourse on immigration in Scandinavia. Quantitative media analyses show that the image of Scandinavian policy divergence is reflected in media discourses on immigration. A content analysis based on the same media data that I apply in a qualitative manner, finds that Danish print media more often mention negative consequences of immigration. Swedish media are more positive, while Norwegian media occupies a position in-between on such a positive-negative continuum (Hovden, Mjelde, & Gripsrud, 2018; see also Greussing & Boomgaarden, 2017;Sakki & Pettersson, 2018). Interestingly, the study also compares Scandinavian broadsheets to broadsheets in France, Germany and the UK, and finds that coverage in all the Scandinavian countries appears to be less problem orientated than in the other countries. This is a reminder that although there are significant differences between the countries, similarities become more pronounced when taking a broader comparative perspective.

Data and methods
The analysis is based on two main types of data: Media texts and policy documents. The reason for combining these two sources is that they together constitute a broader picture of the political context in which the refugee crisis was handled.
I apply a corpus of media texts that was originally collected for the purpose of quantitative content analysis (Hovden et al., 2018), but which is uniquely fitted to this study's most-similar case study design as newspaper articles were specifically sampled to be comparable across the three countries. Trained student assistants collected articles from six Scandinavian newspapers, the largest subscription paper in each country and the largest tabloid: VG, Aftenposten (Norway); Aftonbladet, Dagens Nyheter (Sweden); Ekstrabladet, Jyllandsposten (Denmark). All articles relevant to the refugee crisis were sampled in three different ten-day periods, 1 each characterized by particularly dramatic events, thus presumably generating more coverage: The first mass drownings in the Mediterranean in April, the death of the young boy, Alan Kurdi, in September and the Paris attacks in November. In total this made up a corpus of 304 articles. The periods were chosen on the assumption of this being particularly critical periods with correspondingly high coverage, and because they were international events expected to generate coverage in all countries (and enabling comparison with other countries). Some of the events also coincided with key national events, such as the Swedish turn to restrictive policies in November. An unfortunate effect of this may have been that other key events have evaded sampling, which is why some additional articles were sampled strategically to supplement information on crucial political processes in the three countries. The newspaper articles provide different types of relevant information for the analysis: They give insight into how the refugee crisis was framed differently in the respective national contexts. When politicians are interviewed, they also provide data on policy actors' communicative discourse.
The newspaper articles are supplemented by a selection of relevant policy texts. 2 Using mainly government sources on the internet, I have collected policy documents related to the political agreements that were made in the three countries on special measures to cope with the refugee crisis in the autumn 2015. I have also collected texts on how these agreements were followed up in the following year through actual decisions in the respective parliaments. This includes both commentary and analytical pieces, as well as actual policy documents and debates in national parliaments. The texts were selected in the manner of a "snowball method", starting with commentary pieces (ESPN Flash Reports), following the references to decisions and policy documents given in these, and then further tracing more documents as I gained more insights into the stages of the policy process.
In the analysis of the data I started by going carefully through the newspaper articles, making notes, and coding data using the analytical concepts from discursive institutionalism (cognitive, normative, coordinative, communicative). I also supplemented the corpus by making further searches to follow up on specific issues that appeared in the data. In parallel, I gathered policy texts and secondary literature on policy making in the period, thus making an overview of key developments and significant policy changes in addition to the analytical coding.

Sweden: breaking with humanitarianism?
In a European perspective, the impacts of the refugee crisis in Sweden were dramatic. Only Germany could match Sweden in the number of refugees arriving relative to population size. Ultimately, the strong increase in refugee arrivals led to path-breaking decisions on the regulation of refugee immigration to Sweden, as well as a reorientation of the political discourse on the issue. In the 2018 election campaign, immigration was an important issue. Contrary to both Norway and Denmark however, the strong representation of a right-wing populist anti-immigration party in parliament (since 2010) has not led to any kind of political coalition formation.
Throughout the summer and early autumn of 2015, Swedish authorities took at strong humanitarian position on welcoming the refugees. This is also reflected in the print media, as when an editorial in Aftonbladet says: We cannot solve everything at once. But we can do one thing today. Give the correct answer to the question: Shall we save humans' livesor let them drown? (Aftonbladet 21.4.2015) The question was whether politicians, Swedish and European, were doing enough to assist refugees, and not so much whether they did enough to regulate migration.
As numbers grew, the gap between the number of arrivals and the capacity of the reception apparatus began to appear unmanageable. Reports of refugees who slept under the open sky and of municipalities that struggled to provide schooling, health and social services to swelling numbers of refugees emerged.
The political turning point was dramatically staged. In November 2015 the government called a press conference. Prime minister Stefan Löfven and his deputy Åsa Romson, the latter in tears, announced that they had made the hard decision to restrict entry for new arrivals. The government introduced tightened border controls, and curtailed the residence rights of refugees, shifting from permanent to temporary residence for protection beneficiaries. Furthermore, stricter self-sufficiency and maintenance requirements were introduced for those seeking permanent residence or family reunification. The changes introduced during the fall (some of them before the November press conference) also had a soft side in the shape of measures to improve municipalities' ability to cope with the newcomers. This included extra funds, e.g. for housing projects, but also changes in governance structures where all municipalities were required to settle a certain number of the newcomers and gained new responsibilities for coordinating language and orientation programmes for newcomers (previously the responsibility of the Migration Agency).
None of the changes directly affected the social rights of refugees with legal residency. 3 However, indirectly the introduction of self-sufficiency and maintenance requirements as a condition for obtaining permanent residence or family reunification, means that the residential rights of refugees are made conditional on their ability to succeed in the labour market. Furthermore, when the right to temporary protection expired, the new regulations entitled protection beneficiaries to permanent residency if they had succeeded in obtaining work and thus become self-sufficient. In this sense, the new regulations had a re-commodifying effect for refugees, as their dependence on the market and on employers in order to secure residence for themselves and/or their families, increased.
Some of the changes definitely appeared path-breaking in a Swedish context, although they were moderate if compared to e.g. Danish policies. While both Denmark and Norway had long practiced that refugees were granted temporary protection in the first round, Sweden had in the main granted all protection beneficiaries permanent residency. When this was now reversed, Sweden let go of its image as a humanitarian frontrunner and international exception on immigration policy, and instead accepted that it had to (temporarily at least) lower its standards. How did policy actors legitimize this U-turn discursively?
At the November press conference, the changes were presented as an undesirable, but unavoidable response to extreme pressure. On the front page of Dagens Nyheter, Löfven was quoted «We cannot cope anymore» (Nu klarar vi inte mer). Inside the paper it continued: Prime Minister Stefan Löfven says that the decision has been heavy and painful to make, but that it has been necessary. -We have to act to safeguard that people trust the society and the welfare to work, he says (Dagens Nyheter 25.11.2018).
The new regulations were temporary. They were presented as necessary to acquire "a breather" (andrum) to cope with extraordinary challenges. The government admitted that the restrictions could have negative consequences. At the press conference Åsa Romson called it "a horrible decision which the government feels forced to make" (Dagens Nyheter 25.11.15). Still, the government also maintained that it was necessary that Sweden placed itself at "a minimum level according to EU-law and international conventions" in order to "gain breathing space for Swedish reception of refugees, more of EU's member states must take their [share of the] responsibility" (Swedish government, 2016, p. 30). This was grounded in an analysis that the favourable conditions offered to asylum-seekers in Sweden led fewer refugees to seek asylum in other member states, thus making it easier for other states to dodge their share of the burden (Swedish government, 2016, p. 30). In this way of reasoning, the point was not so much to discourage refugees from coming to Sweden, as to put pressure on other states to take their part of the responsibility.
The most prominent cognitive arguments for the changes were thus grounded in a causal narrative about asylum magnets, according to which favourable conditions in receiving countries work to attract more asylum seekers to the countries that offer good conditions and thus allow countries with restrictive policies to dodge their rightful part of the responsibility. In other words, to promote the goal of a more equitable and fair sharing of the burden of receiving refugees, Sweden could not offer better conditions and prospects than other countries.
Another cognitive argument for the necessity of changes was based in the analysis of overstretched capacities in the Swedish welfare system. In the media, reports about refugees sleeping rough and about the burst capacities of local education and health services emerged. Aftonbladet's commentator Lena Melin wrote: «Life and health and vital social functions are at stake following the refugee disaster» (Aftonbladet 19.11.2015). This is very close to the argument of the prime minister, as quoted above, legitimising the government's u-turn by assuring that they were acting to safeguard people's trust in society and the welfare system.
Normatively, the authorities' argumentation appealed to values of international collaboration and shared responsibility, as well as the protection of established institutions for protecting refugees' rights. Policy actors emphasised that none of the restrictions went beyond the limits established by EU-law and international conventions. Furthermore, they were motivated by the need to make all European countries take their share of the responsibility for the world's refugeesto promote international solidarity and collaboration.
There is an emphasis on both urgency and continuity in the government's efforts to communicate the legitimacy of the changes. The changes respond to a dramatic situation, they are ultimately undesirable, but yet necessary given the urgency of the problem. On the other hand, the changes do not represent a change of heart, merely of methods. The values appealed to are the same values of international solidarity, institutionalized rights for refugees and humanitarian ideals that would also motivate a liberal refugee policy.
The press coverage paid little attention to efforts at constructing a coordinative discourse between policy actors to cope with the refugee crisis. Instead, it seems like the most pressing concern was to create legitimacy for the policy changes in the wider public. The sense of urgency created by the refugee crisis was sufficient to establish political agreement on new measures. The challenge was rather to establish a normative foundation for a reorientation of policy.
The temporary measures chosen in an extreme moment in time has so far proved to be of a more permanent character. Three years later, the new redgreen government declared an extension of the temporary measures from 2015, but with extension of rights to family reunification also for non-convention protectionaries. The argumentation was the same: "More countries have to take a greater responsibility. Swedish reception of refugees have to be sustainable longterm" (Löfven, 2019). Also in the area of integration, the government declaration shows signs of on-going path change: Enforced activation measures and, not least, the introduction of compulsory language and social knowledge requirements for citizenshipthus making an end to Swedish exceptionalism in yet another field (Midtbøen, 2015).

Denmark: continuing along its own path
In Denmark, comparatively restrictive policies were already in place to minimize the impact of international refugee movements. The country has become internationally notorious for its harsh stance on immigration. Despite this, there was a steep increase in the number of asylum-seekers also in Denmark during 2015, although both the increase and the absolute figures were moderate compared to its neighbours. Indeed the striking images of hundreds of refugees wandering along Danish highways in September 2015, were reportedly pictures of people who did not want to come to Denmark. They were transiting on their way to Sweden, a country with a far more favourable reputation for welcoming refugees.
Already in the coverage of the spring mass drownings in the Mediterranean, differences between Danish and Swedish news reporting become clear. First, the very problem construction differ. Where the Swedish newspapers criticize European politicians for not helping, a Danish editorial construes the liberality of Europe's refugee policy as part of the problem: Europe's politicians are co-responsible for what is going on, because they maintain the perspective that refugees from poverty can succeed in reaching the promised continent. They have not been able tighten immigration legislation in a manner where it would be clear even to the last human smuggler that you cannot get residence in Europe purely for economic reasons. In this way, the EU's political elite takes part in maintaining the Mediterranean as a channel of death, as it has been so grotesquely expressed. (Jyllandsposten 21.4.2015) Second, in the early stages the Swedish papers tended to portray the refugee crisis primarily as an international tragedy where all countries had to do their share to solve a humanitarian problem. In the Danish newspapers, the crisis is debated in terms of implications for Danish policy: how can Danish policy makers act to minimize the impacts of the refugee crisis on Denmark? Different opinions were voiced on how to do this, but the point is that the refugee crisis at an early stagelong before the refugees made a significant physical presence in Denmarkwas construed as a potentially national problem.
As in Sweden, the Danish government presented a range of new measures designed to deter new arrivals and to smooth the reception of those who had come. Many such measures were in fact launched prior to the events that were termed the refugee crisis. During the winter, the social democratic government tightened asylum regulations by restricting access to permanency and family reunification for some types of protection beneficiaries. In June, this government lost the general elections. It was replaced by a conservative-liberal coalition, supported by a strong Danish People's Party. The new government immediately set out to introduce a number of restrictive policies including lower social benefits for asylum-seekers and immigrants, stricter requirements for permanent residency for refugees (including language and self-sufficiency) and stricter asylum controls.
A new and lower integration benefit, which replaced social assistance for those who have not been in Denmark for more than seven of the last 8 years, was introduced already in September 2015. In November, the government announced a 34-items asylum policy package adopted by a parliamentary majority (including the social democrats) in November and January 2016 (Danish Government, 2015). Most controversy surrounded the so-called "jewellery regulation". As refugees were expected to support themselves when possible, the police was authorized to confiscate refugees' valuables exceeding €1340 without sentimental value (Kvist, 2016). Since 2016 further changes have also been made to the social rights of newly arrived migrants. A work requirement was introduced as a condition to receive the integration benefit. Following tripartite negotiations (between government, municipalities and the social partners) a new Integration Education Benefit was established, which implies that refugees are allocated to municipalities where their qualifications match labour market demand and where they work in parallel to learning Danish etc. (Kvist, 2016).
In the Danish news coverage I have examined, there are few articles in which politicians explicitly argue for the changes. Instead, most articles are about the political negotiations between parties to secure parliamentary support for the government's proposals. It seems like the coordinating discourse that went on outside of the public eye was more interesting for the Danish papers than the specific arguments politicians used to legitimize their positions to the public. This may reflect that Danish policy actors did not have the same need to secure legitimacy for stricter policies through communicative discourse to the public, as a general legitimacy for restrictive policies was already established.
To identify the Danish government's normative and cognitive arguments for the changes I have instead turned to the policy documents in which the proposals were presented at the government's website. This is how the government introduces its "asylum policy package": There are many refugees coming to Europe. This is putting pressure on all countries also Denmark. And the pressure rises day by day. We are taking on a responsibility. But we shall not receive so many refugees, that it will threaten the social cohesion of our own country. The number [of refugees] matter for whether integration can succeed. And it means something, for whether we can maintain a good and secure society. The Government has long been of the opinion that the requirements in the asylum area means something for how attractive it is to come to Denmark. That is why the Government introducedright in the beginninga new and lower integration allowance. And that this why the Government now wants to tighten conditions even more for asylum requirements and entry to Denmark. (Danish government, 2015).
There is a story about numbers and carrying capacity in this argument. A cognitive argument states that there is a limit to how many refugees Demark can receive. Too many will threaten the social cohesion of the entire society. The limit is unspecified, but as pressure is rising, it is urgent to take steps to limit numbers irrespective of exact figures.
There are similarities between this story and the Swedish government's story about the perils of offering newcomers access to residence and social rights. They both presuppose a causal effect from conditions offered to arrival figures. However, where Swedish policy changes are about not offering better conditions than other countries, the Danish government portrays itself as a frontrunner in reducing the entitlements of asylum-seekers and refugees. There is no path-breaking in the Danish government's self-presentation, it is continuing along an established path of restrictive policies. Thus there is no need to apologize and justify the changes in the way that Swedish politicians did. On the contrary, there is reassurance of continuity.
What values does this policy proposal appeal to? Primarily Danish national interests. The normative appeal is to the basis of community in the shape of a national "we". There is no reference to international solidarity or conventions, but to the desire to protect a "good and secure society", "our own country" (Danish government, 2015:1). It is the security of this national we which is at stake, not the refugees' need for protection.
While it is possible to detect a similar goal and similar cognitive content across Swedish and Danish policies in the sense that both seek to reduce the number of refugees arriving, they appeal to highly discrepant values when arguing for these goals. While Denmark introduced more far-reaching changes to new migrants' social benefits than both Sweden and Norway, only limited efforts have been put into establishing a communicative discourse to legitimize the changes. As the changes has continued a path of restrictive policies established already in the late 1990s (Jønsson & Petersen, 2012), this has not been necessary. Instead, precisely because the proposed policy instruments were specific and numerous, the challenge has been to establish a coordinating discourse in order to secure broad and stable political support for concrete measures, including restrictions in refugees' social rights.

Norway: still in the middle
The early coverage of the refugee crisis in the Norwegian newspapers was characterised by a strong emphasis on what actions Norway, as a distinct actor, should take to relieve a crisis that was predominantly described as taking place somewhere else. There are calls for sending Norwegian ships to assist in the rescue efforts in the Mediterranean. Furthermore, opposition parties rallied around a call for an extraordinary quota of 10.000 refugees. Headlines such as Syria dugnad (which approximately can be translated into collective voluntary work for Syria) were used, thus appealing to citizens' obligations to assist (Aftenposten 22.04.2015). The right-wing government was critical, drawing on cognitive arguments on how resources should be most efficiently spent: There are millions of refugees, and four great, humanitarian disasters are ongoing in the world at the same time. In this situation, the least efficient measure is to bring people her, compared with using money to aid them in close areas, says the Prime Minister in the programme. (Aftenposten 22.04.2015) Later in the spring, a parliamentary majority settled on the figure 8000 quota refugees. The decision was controversial but testifies that the implications of the crisis in terms of reception and integration potentials in Norway, were on the agenda already in the springat a time when the crisis predominantly were taking place in a physically distant location.
This changed in the fall. Striking pictures of refugees who cycled across the Norwegian-Russsian border at Storskog began to appear in the press. Russian border regulations meant that border crossings were not allowed on foot, and refugees arrived on cheap bikes that piled up at the Norwegian side. The little heard of northern border post suddenly became the focal point of the Norwegian press coverage of the refugee crisis. Nearly 5500 asylum-seekers crossed the border at Storskog in 2015. In this situation, the need to reduce the number of arrivals became urgent, both for the government and the opposition, and a process of establishing a consensus on how to deal with the crisis commenced.
An asylum agreement between the government parties (Conservative Party and Progress Party), the centre (Liberals and Christian Democrats), the Centre Party and the Labour party was reached in November. Political actors referred to the urgency of the situation: the situation was difficult, it was extraordinary, people needed to know that borders were under control (Aftenposten 20.11.2015).
The agreement was inward-looking tending to the immediate situation in Norway (Brekke & Staver, 2018). It included faster return and other measures to deter asylumseekers, but also several clauses affecting the social rights of asylum-seekers and refugees. It obliged the government to «Ensure that asylum seekers' social benefits are on a level where Norway does not appear as economically attractive in relation to comparable European countries» (Norwegian Government, 2015, item 6). This could be interpreted as mainly regarding the immediate level of benefits asylum seekers received. But the agreement also contained clauses that concerned social rights for protection beneficiaries with long-term or permanent residence. This included longer waiting times before qualifying for family benefits and a call for the government to «review special measure for refugees and asylum seekers in the social security legislation and make sure that activity-and contribution requirements are made when natural» (Norwegian government, 2015, item 17). In this sense, a coordinating discourse seemed to have been established for reforming refugees' social rights in significant ways.
As a follow-up, the government presented an audit document with a long list of proposals to tighten immigration regulation. This faced serious criticism, also from the parties that had partaken in the asylum agreement of November. Attempts by the Conservative Prime Minister to negotiate a broad compromise failed as the Progress Party insisted on having separate votes on each proposal in the package. This allowed the party to demonstrate its restrictive position and thus also reconfirm its image as hardliners on immigration despite being part of a more moderate government. However, it also meant that several of the most controversial proposals fell (Pedersen, 2016). In short, a narrow communicative discourse with the Progress Party's own constituency appears to have been prioritised at the cost of coordinating discourse with other policy actors. Still, even though the most radical proposals were rejected, new restrictive measures were also adopted. This included stricter integration requirements for permanent residency (self-sufficiency and passed tests in Norwegian language etc.) and stricter requirements on age and belonging to Norway for family reunification (Pedersen, 2016).
In a second audit document from July 2016, the government made further proposals to limit the social rights of protection beneficiaries. Again, the proposals were quite radical and pointing towards a stronger dualisation of social security (Pedersen, 2017). The first set of proposals involved extending waiting periods before immigrants are eligible for social security benefits such as old age pension, disability pension, rehabilitation benefits and lone parents benefits. The second set of proposals would remove all existing preferential treatment of refugees compared to other migrant groups. Convention refugees have, for example, been exempted from waiting periods and their pension entitlements compensated for short residency times.
Despite the previous coordinated call for the government to "make sure that the level of benefits for asylum-seekers should be of a kind that does not make Norway appear financially attractive in relation to comparable European countries" (Norwegian government, 2015), the governing parties found hardly no support in the parliament for its new proposals. The narrative where it was key not to appear attractive, not to become a welfare magnet, had at this point apparently been left by all parties. The very problem construction had changed from one of acute need for migration control to a more conventional welfare state problem of activation. The government parties appealed to the need to give refugees work incentives. Against this, the opposition warned against growing poverty as an effect of the proposed welfare cuts. In other words, the framing of refugees' social rights as an element of immigration control policy, which had come to the fore during the "crisis", had been abandoned. Instead, the issue of refugees' social benefits had returned to a primarily national welfare state orientated discursive framework where concerns about work incentives and poverty were pitted against each other (much as is the case with general debates over all other social benefit schemes). In this sense, the Progress Party had lost a potential "window of opportunity" to move Norwegian policy further in a Danish direction.

Discussion
There is little doubt that the refugee crisis of 2015 was experienced as a dramatic event in both Sweden, Denmark and Norway, and that it sparked off processes of policy change in a more restrictive direction. At the same time, the crisis was construed in rather different ways in the three respective public spheres, and the types of changes that resulted were, as suggested by previous research, discrepant and shaped by different national contexts.
The political discourse in the three countries positions them very differently with respect to the part the states play internationally. Sweden is portrayed as at heart a frontrunner on humanitarian immigration policies. Restrictive change is legitimised as a move of international collaboration, stimulating more countries to take their share of the responsibilityquite the opposite of a "race to the bottom". Danish political actors insist on continuing an already restrictive line. Denmark is portrayed as a frontrunner in the opposite direction, actively working to not appear as an attractive or soft target for international migrants and refugeesit wants to lead the "race to the bottom". The Norwegian ambition seems to be to excel neither on the restrictive nor liberal side of the spectrum, but to do its share. The notion of dugnadvolunteer workappears in several contexts in the material. Local involvement and engagement are cherishedwhether in the shape of sending ships to rescue refugees in the Mediterranean Sea, or in the case of hotel mogul Petter Stordalen welcoming refugees to stay in his hotel (VG 08.09.2016). The country is no frontrunner in either direction, but it recognizes a moral obligation to take part in a collective effort. This rhetoric of dugnad and shared efforts seem to have been efficient, at least for a time, in coordinating actors across political divides.
It is an interesting finding that immigration and integration political institutions seem to differ with respect to their resilience. Policies related to border control and residence rights were more susceptible to change than those related to the social rights of migrants, especially in Sweden and Norway. One reason may be that these are measures closely tied up to general welfare state institutions, such as family policy, social assistance and principles of universal and equal rights for all residents. As such they mobilised conventional left-right divides in a way that border control policies do not. In Denmark it was easier to muster cross-political support for such changes, as the pathbreaking step towards welfare dualism had already been taken many years before the events of 2015, when an integration benefit with lower support levels for immigrants was first adopted in 1999 (Jønsson & Petersen, 2012). However, also in Norway signs of an emerging coalition on reforming refugees social rights were strong in the early stages, but the government failed in sustaining a coordinative discourse over time.
The relative significance of communicative versus coordinative discourse in the three countries vary considerably. In previous research, varying significance of coordinative and communicative discourse have been linked to differences in political institutions (Schmidt, 2002) and to different stages in the policy process (Ervik & Lindén, 2015). This study indicates that also the political sensitivity of the matter at stake is significant. In Sweden, the sense of crisis and urgency seemed to make the need for coordination less. At the same time, the need to legitimize immigration political change was acute, and great efforts were placed on providing a normative defence of the changes that maintained the country's humanitarian identity. This work is by no means finished, as the country's political elite is struggling to come to terms with the growth of the Sweden Democrats. In Denmark, there was limited need for extra efforts with respect to legitimizing communicative discourse. Instead, the media focused on political actors' work to establish a coordinative discourse to secure broad support for a range of new concrete measuresalso on the social policy side. Coordination with respect to restrictive immigration policies is strong, but on the more controversial issue of lower social assistance for newcomers, a left-right divide still exists (as reflected in social democratic governments' reversal of some policies). Again, Norway seems to adopt a more mixed position in the middle. A period of efficient coordination was aborted, as the governing Progress Party prioritised communicative discourse to its antiimmigration heartland at the expense of coordination in the parliament. The shared language of urgency instigated by the crisis, eventually lost its power, and the parties returned to their old trenches. Crises can certainly open up windows of opportunity for radical policy change, but this requires intensive discursive work by policy actors in creating popular legitimacy and sufficiently broad political support to sustain the changes over time. The functions of different types of discursive efforts in different types of policy change should be explored in future research. One hypothesis derived from this study is that a change of overall policy goals require communicative discourse, while the transformation of specific policy instruments rely more on efficient coordinative discourse between policy actors.